Background-Nucleotide variants in several genes for lipid and methionine metabolism influence the risk of premature atherosclerosis. Ten percent of single nucleotide substitutions in these genes involve mRNA splice sites. The effects of some of these changes on splicing and on phenotypic severity are not inherently obvious. Methods and Results-Using an information theory-based model, we measured the individual information content (R i , in bits) of splice sites adjacent to 289 mutations (including 31 splice-site mutations) in the atherosclerosis candidate genes APOAII, APOB, APOCII, APOE, CBS, CETP, LCAT, LIPA, LDLR, and LPL. The predictions of information analysis were then corroborated by published mRNA analyses. The R i values of mutant sites were consistent with either complete (nϭ17) or partial (nϭ8) inactivation of these sites. Seven mutations were predicted to activate cryptic splice sites. Predicted inactive mutant sites were associated with either "average" or "severe" dyslipidemia and commensurate reductions in protein levels or activity, whereas mutations expected to exhibit residual splicing had average or "mild" effects on lipid and protein expression. Conclusions-Information analysis of splice-junction variants in atherosclerosis candidate genes distinguishes inactive from leaky splice sites and identifies activated cryptic sites. Predicted changes in splicing were related to phenotypic severity. (Circulation. 1999;100:693-699.)